Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
2. The method of claim 1 , wherein: the random matrix includes: a size including a raw template dimension by a transformed template dimension; and values of 0 or 1, wherein a probability of a particular element of the random matrix being 1 is a particular probability.
In the privacy-preserving biometric authentication method where a random projection transforms raw biometric data, the random matrix used for the transformation is a matrix containing elements that are either 0 or 1. The matrix has dimensions of 'raw template dimension' (size of the input biometric) by 'transformed template dimension' (size of the privacy-protected biometric). Each element in this matrix has a specific probability of being a 1.
3. The method of claim 1 , wherein the first transformed biometric template and the second transformed biometric template are members of a vector space having a transformed template dimension.
In the privacy-preserving biometric authentication method where a random projection transforms raw biometric data, the transformed biometric templates (both the registration template and the challenge template) exist within a vector space. This vector space has a dimensionality equal to the 'transformed template dimension', defining the space where these privacy-protected biometric representations reside.
4. The method of claim 1 , wherein the approximate matching is based on a hamming distance between the first transformed biometric template and the second transformed template.
In the privacy-preserving biometric authentication method where a random projection transforms raw biometric data, the method determines how well the registration biometric matches the challenge biometric by calculating the Hamming distance between their transformed biometric templates. A smaller Hamming distance indicates a better match.
5. The method of claim 1 , further comprising receiving initialization data including a first hamming value, a second hamming value, and a transformed hamming threshold, wherein: the first hamming value is less than the second hamming value; the second hamming value is less than a transformed template dimension; when a hamming distance between the first raw biometric template and the second raw biometric template less than the first hamming value, then a probability of a hamming distance between the first transformed biometric template and the second transformed biometric template being greater than the transformed hamming threshold is less than 1 in 10,000; and when the hamming distance between the first raw biometric template and the second raw biometric template is greater than the second hamming value, then a probability of the transformed hamming distance between the first transformed biometric template and the second transformed biometric template being less than or equal to the transformed hamming threshold is less than 1 in 10,000.
In the privacy-preserving biometric authentication method using random projections, the process includes receiving initialization data: `first_hamming_value`, `second_hamming_value`, and `transformed_hamming_threshold`. The `first_hamming_value` is smaller than `second_hamming_value`, which is smaller than the `transformed_template_dimension`. If the Hamming distance between raw biometrics is less than `first_hamming_value`, the chance that the transformed Hamming distance exceeds `transformed_hamming_threshold` is very low (less than 1/10,000). Conversely, if the raw Hamming distance is greater than `second_hamming_value`, the chance of the transformed Hamming distance being less than or equal to `transformed_hamming_threshold` is also very low (less than 1/10,000). These thresholds help to determine authentication.
6. The method of claim 1 , wherein: the first raw biometric template is defined according to an expression X T ε{0,1} n ; the second raw biometric template is defined according to an expression Y T ε{0,1} n ; the first transformed biometric template is defined according to an expression X T′ =RX T ε{0,1} k ; the second transformed biometric template is defined according to an expression Y T′ =RY T ε{0,1} k , in which: Y T represents the second raw biometric template; {0,1} k represents a vector space having a dimension that includes the first transformed template dimension; and Y T′ represents the second transformed biometric template.
In the privacy-preserving biometric authentication method, the raw biometric templates are bit strings. The first raw template (`XT`) is represented as a vector of length `n` with elements that are 0 or 1. The second raw template (`YT`) also follows the same format. After the random projection (R), these become `XT'` and `YT'`, respectively. These transformed templates reside in a vector space of dimension `k`, which represents the transformed template dimension. The transformation is defined by `XT' = R * XT` and `YT' = R * YT`, with `YT'` representing the transformed version of the second raw biometric.
8. The non-transitory computer-readable medium of claim 7 , wherein: the random matrix includes: a size including a raw template dimension by a transformed template dimension; and values of 0 or 1, wherein a probability of a particular element of the random matrix being 1 is a particular probability.
In the privacy-preserving biometric authentication method where a random projection transforms raw biometric data and is implemented on a non-transitory computer-readable medium, the random matrix used for the transformation is a matrix containing elements that are either 0 or 1. The matrix has dimensions of 'raw template dimension' (size of the input biometric) by 'transformed template dimension' (size of the privacy-protected biometric). Each element in this matrix has a specific probability of being a 1.
9. The non-transitory computer-readable medium of claim 7 , wherein the first transformed biometric template and the second transformed biometric template are members of a vector space having a transformed template dimension.
In the privacy-preserving biometric authentication method where a random projection transforms raw biometric data and is implemented on a non-transitory computer-readable medium, the transformed biometric templates (both the registration template and the challenge template) exist within a vector space. This vector space has a dimensionality equal to the 'transformed template dimension', defining the space where these privacy-protected biometric representations reside.
10. The non-transitory computer-readable medium of claim 7 , wherein the approximate matching is based on a hamming distance between the first transformed biometric template and the second transformed template.
In the privacy-preserving biometric authentication method where a random projection transforms raw biometric data and is implemented on a non-transitory computer-readable medium, the method determines how well the registration biometric matches the challenge biometric by calculating the Hamming distance between their transformed biometric templates. A smaller Hamming distance indicates a better match.
11. The non-transitory computer-readable medium of claim 7 , wherein: the operations further comprise receiving initialization data including a first hamming value, a second hamming value, and a transformed hamming threshold; the first hamming value is less than the second hamming value; the second hamming value is less than a transformed template dimension; when a hamming distance between the first raw biometric template and the second raw biometric template less than the first hamming value, then a probability of a hamming distance between the first transformed biometric template and the second transformed biometric template being greater than the transformed hamming threshold is less than 1 in 10,000; and when the hamming distance between the first raw biometric template and the second raw biometric template is greater than the second hamming value, then a probability of the transformed hamming distance between the first transformed biometric template and the second transformed biometric template being less than or equal to the transformed hamming threshold is less than 1 in 10,000.
In the privacy-preserving biometric authentication method using random projections and implemented on a non-transitory computer-readable medium, the process includes receiving initialization data: `first_hamming_value`, `second_hamming_value`, and `transformed_hamming_threshold`. The `first_hamming_value` is smaller than `second_hamming_value`, which is smaller than the `transformed_template_dimension`. If the Hamming distance between raw biometrics is less than `first_hamming_value`, the chance that the transformed Hamming distance exceeds `transformed_hamming_threshold` is very low (less than 1/10,000). Conversely, if the raw Hamming distance is greater than `second_hamming_value`, the chance of the transformed Hamming distance being less than or equal to `transformed_hamming_threshold` is also very low (less than 1/10,000). These thresholds help to determine authentication.
12. The non-transitory computer-readable medium of claim 7 , wherein: the first raw biometric template is defined according to an expression X T ε{0,1} n ; the second raw biometric template is defined according to an expression Y T ε{0,1} n ; the first transformed biometric template is defined according to an expression X T′ =RX T ε{0,1} k ; the second transformed biometric template is defined according to an expression Y T′ =RY T ε{0,1} k , in which: Y T represents the second raw biometric template; {0,1} k represents a vector space having a dimension that includes the first transformed template dimension; and Y T′ represents the second transformed biometric template.
In the privacy-preserving biometric authentication method implemented on a non-transitory computer-readable medium, the raw biometric templates are bit strings. The first raw template (`XT`) is represented as a vector of length `n` with elements that are 0 or 1. The second raw template (`YT`) also follows the same format. After the random projection (R), these become `XT'` and `YT'`, respectively. These transformed templates reside in a vector space of dimension `k`, which represents the transformed template dimension. The transformation is defined by `XT' = R * XT` and `YT' = R * YT`, with `YT'` representing the transformed version of the second raw biometric.
14. The method of claim 13 , wherein the random matrix includes: a size including a raw template dimension by a transformed template dimension; and values of 0 or 1, wherein a probability of a particular element of the random matrix being 1 is a particular probability.
In the privacy-preserving biometric authentication method performed by an authentication server where a random projection transforms raw biometric data, the random matrix used for the transformation is a matrix containing elements that are either 0 or 1. The matrix has dimensions of 'raw template dimension' (size of the input biometric) by 'transformed template dimension' (size of the privacy-protected biometric). Each element in this matrix has a specific probability of being a 1.
15. The method of claim 13 , further comprising: in response to the hamming distance being above the transformed hamming threshold, not communicating the authentication signal to the user device.
In the privacy-preserving biometric authentication method performed by an authentication server, where a random projection transforms raw biometric data and the transformed templates are compared, if the Hamming distance between the transformed biometric templates is greater than the pre-defined `transformed_hamming_threshold`, the authentication server will *not* send the authentication success/failure signal back to the user's device. This hides information about the comparison result when there is not a good match.
16. The method of claim 13 , wherein: the first transformed biometric template and the second transformed biometric template are members of a vector space having a transformed template dimension.
In the privacy-preserving biometric authentication method performed by an authentication server where a random projection transforms raw biometric data, the transformed biometric templates (both the registration template and the challenge template) exist within a vector space. This vector space has a dimensionality equal to the 'transformed template dimension', defining the space where these privacy-protected biometric representations reside.
17. The method of claim 13 , further comprising selecting the transformed hamming threshold and the initialization data including a first hamming value, a second hamming value, and a transformed hamming threshold, wherein: the first hamming value is less than the second hamming value; the second hamming value is less than a transformed template dimension; when a hamming distance between the first raw biometric template and the second raw biometric template less than the first hamming value, then a probability of a hamming distance between the first transformed biometric template and the second transformed biometric template being greater than the transformed hamming threshold is less than 1 in 10,000; and when the hamming distance between the first raw biometric template and the second raw biometric template is greater than the second hamming value, then a probability of the transformed hamming distance between the first transformed biometric template and the second transformed biometric template being less than or equal to the transformed hamming threshold is less than 1 in 10,000.
In the privacy-preserving biometric authentication method performed by an authentication server, the server selects the `transformed_hamming_threshold` and initialization data (`first_hamming_value`, `second_hamming_value`). `first_hamming_value` is less than `second_hamming_value`, which is less than the `transformed_template_dimension`. If the Hamming distance between raw biometrics is less than `first_hamming_value`, the probability that the transformed Hamming distance exceeds `transformed_hamming_threshold` is low (less than 1/10,000). Conversely, if the raw Hamming distance exceeds `second_hamming_value`, the probability of the transformed Hamming distance being less than or equal to `transformed_hamming_threshold` is also low (less than 1/10,000). This threshold selection ensures privacy and accuracy.
18. The method of claim 13 , wherein: the first raw biometric template is defined according to an expression X T ε{0,1} n ; the second raw biometric template is defined according to an expression Y T ε{0,1} n ; the first transformed biometric template is defined according to an expression X T′ =RX T ε{0,1} k ; the second transformed biometric template is defined according to an expression Y T′ =RY T ε{0,1} k , in which: Y T represents the second raw biometric template; {0,1} k represents a vector space having a dimension that includes the first transformed template dimension; and Y T′ represents the second transformed biometric template.
In the privacy-preserving biometric authentication method performed by an authentication server, the raw biometric templates are bit strings. The first raw template (`XT`) is represented as a vector of length `n` with elements that are 0 or 1. The second raw template (`YT`) also follows the same format. After the random projection (R), these become `XT'` and `YT'`, respectively. These transformed templates reside in a vector space of dimension `k`, which represents the transformed template dimension. The transformation is defined by `XT' = R * XT` and `YT' = R * YT`, with `YT'` representing the transformed version of the second raw biometric.
19. The method of claim 13 , further comprising encrypting the registration template.
In the privacy-preserving biometric authentication method performed by an authentication server that uses random projections, the registration template (the transformed biometric data sent to the server during enrollment) is encrypted to further protect user data at rest.
21. The non-transitory computer-readable medium of claim 20 , wherein the random matrix includes: a size including a raw template dimension by a transformed template dimension; and values of 0 or 1, wherein a probability of a particular element of the random matrix being 1 is a particular probability.
In the privacy-preserving biometric authentication method performed by an authentication server and implemented on a non-transitory computer-readable medium where a random projection transforms raw biometric data, the random matrix used for the transformation is a matrix containing elements that are either 0 or 1. The matrix has dimensions of 'raw template dimension' (size of the input biometric) by 'transformed template dimension' (size of the privacy-protected biometric). Each element in this matrix has a specific probability of being a 1.
22. The non-transitory computer-readable medium of claim 20 , wherein the operations further comprise: in response to the hamming distance being above the transformed hamming threshold, not communicating the authentication signal to the user device.
In the privacy-preserving biometric authentication method performed by an authentication server, implemented on a non-transitory computer-readable medium, where a random projection transforms raw biometric data and the transformed templates are compared, if the Hamming distance between the transformed biometric templates is greater than the pre-defined `transformed_hamming_threshold`, the authentication server will *not* send the authentication success/failure signal back to the user's device. This hides information about the comparison result when there is not a good match.
23. The non-transitory computer-readable medium of claim 20 , wherein: the first transformed biometric template and the second transformed biometric template are members of a vector space having a transformed template dimension.
In the privacy-preserving biometric authentication method performed by an authentication server and implemented on a non-transitory computer-readable medium where a random projection transforms raw biometric data, the transformed biometric templates (both the registration template and the challenge template) exist within a vector space. This vector space has a dimensionality equal to the 'transformed template dimension', defining the space where these privacy-protected biometric representations reside.
24. The non-transitory computer-readable medium of claim 20 , wherein: the operations further comprise selecting the transformed hamming threshold and the initialization data including a first hamming value and a second hamming value; the first hamming value is less than the second hamming value; the second hamming value is less than a transformed template dimension; when a hamming distance between the first raw biometric template and the second raw biometric template less than the first hamming value, then a probability of a hamming distance between the first transformed biometric template and the second transformed biometric template being greater than the transformed hamming threshold is less than 1 in 10,000; and when the hamming distance between the first raw biometric template and the second raw biometric template is greater than the second hamming value, then a probability of the transformed hamming distance between the first transformed biometric template and the second transformed biometric template being less than or equal to the transformed hamming threshold is less than 1 in 10,000.
In the privacy-preserving biometric authentication method performed by an authentication server and implemented on a non-transitory computer-readable medium, the server selects the `transformed_hamming_threshold` and initialization data (`first_hamming_value`, `second_hamming_value`). `first_hamming_value` is less than `second_hamming_value`, which is less than the `transformed_template_dimension`. If the Hamming distance between raw biometrics is less than `first_hamming_value`, the probability that the transformed Hamming distance exceeds `transformed_hamming_threshold` is low (less than 1/10,000). Conversely, if the raw Hamming distance exceeds `second_hamming_value`, the probability of the transformed Hamming distance being less than or equal to `transformed_hamming_threshold` is also low (less than 1/10,000). This threshold selection ensures privacy and accuracy.
25. The non-transitory computer-readable medium of claim 20 , wherein: the first raw biometric template is defined according to an expression X T ε{0,1} n ; the second raw biometric template is defined according to an expression Y T ε{0,1} n ; the first transformed biometric template is defined according to an expression X T′ =RX T ε{0,1} k ; the second transformed biometric template is defined according to an expression Y T′ =RY T ε{0,1} k , in which: Y T represents the second raw biometric template; {0,1} k represents a vector space having a dimension that includes the first transformed template dimension; and Y T′ represents the second transformed biometric template.
In the privacy-preserving biometric authentication method performed by an authentication server and implemented on a non-transitory computer-readable medium, the raw biometric templates are bit strings. The first raw template (`XT`) is represented as a vector of length `n` with elements that are 0 or 1. The second raw template (`YT`) also follows the same format. After the random projection (R), these become `XT'` and `YT'`, respectively. These transformed templates reside in a vector space of dimension `k`, which represents the transformed template dimension. The transformation is defined by `XT' = R * XT` and `YT' = R * YT`, with `YT'` representing the transformed version of the second raw biometric.
26. The non-transitory computer-readable medium of claim 20 , wherein the operations further comprise encrypting the registration template.
In the privacy-preserving biometric authentication method performed by an authentication server and implemented on a non-transitory computer-readable medium that uses random projections, the registration template (the transformed biometric data sent to the server during enrollment) is encrypted to further protect user data at rest.
Unknown
September 26, 2017
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.